A 15% Yield on AI Debt—That’s the Market Asking Hard Questions

CoreWeave’s bonds are yielding around 15%. Let that number sink in for a second, because it’s doing a lot of storytelling here. 🚨

When a company borrows money by issuing bonds, the yield is basically the return lenders demand for taking on the risk that the company might not pay them back. Ultra-safe borrowers pay low yields. Risky borrowers have to sweeten the pot. So a 15% yield means the bond market is looking at CoreWeave and saying: “Sure, we’ll lend you money… for a price that assumes things could get weird.”

Stock investors see a hypergrowth AI story. Bond investors see a leveraged one. Both can be right at the same time — which is exactly why this is worth breaking down.

## What actually happened

CoreWeave (ticker CRWV) is one of the fastest-growing companies in tech. At the end of 2022, it operated three data centers. Today it runs 51, packed with more than a quarter-million Nvidia GPUs — the chips that power AI training and inference. Its revenue went from $16 million in 2022 to $5.13 billion in 2025. That’s not a typo. That’s one of the most explosive revenue ramps in recent memory.

The reason is the AI boom. Companies that don’t want to build their own AI data centers rent computing power from “neoclouds” like CoreWeave — essentially specialized cloud providers built specifically for AI workloads, as opposed to the general-purpose clouds run by Amazon, Microsoft and Google.

But here’s the catch, and it’s a big one: GPUs are absurdly expensive. Data centers are absurdly expensive. CoreWeave didn’t have $16 million-to-$5 billion growth funded out of its pocket. That expansion has been financed largely with debt — and the bond market is now charging CoreWeave a steep interest rate to keep the machine running.

## Why the 15% yield matters

Think of it this way: if your friend with a shaky job history asked to borrow $100 and offered you 15% interest, you’d take the deal but also quietly worry about them. That’s the bond market’s vibe right now.

A high yield does two things to a company:

First, it makes future borrowing expensive. Every new data center, every new GPU order, gets financed at worse rates. Growth gets more costly.

Second, it creates a stack of interest payments that come due no matter what. Stockholders get paid after everyone else. If revenue growth slows — even a little — those interest payments start eating the cash flow that was supposed to fund the next phase of expansion.

This is the core tension with CoreWeave: the growth is real, but the growth is bought. It’s a race to build enough capacity to lock in long-term AI contracts before the debt load becomes a problem.

## The trade-off investors need to see

Revenue growing from $16 million to $5.13 billion in three years is genuinely elite. But revenue alone doesn’t tell you whether the business works. The questions that matter here:

– Are those AI contracts locked in for years, or do they concentrate among a few big customers who could walk?
– Is the company actually generating free cash flow — money left over after paying for GPUs, data centers and interest — or is it borrowing more to cover the bills?
– What happens if AI demand cools even 20%? The debt payments don’t cool with it.

There’s also a quieter risk people skip: technology. CoreWeave is buying today’s top Nvidia chips at today’s prices. If next-generation chips make current ones less valuable, or if the big cloud providers build their own AI infrastructure at scale, the economics of leasing GPUs could compress.

None of this means CoreWeave is doomed. It means the bond market is pricing in real uncertainty, and stock investors should at least know that conversation is happening.

## What’s signal, what’s hype?

Signal: explosive revenue growth, massive infrastructure buildout, genuine demand for AI computing capacity.

Hype risk: assuming the growth rate continues forever, and treating the stock price as proof the balance sheet is fine. It isn’t proof. A stock can rally hard while the company racks up expensive obligations underneath. Green candles have convinced plenty of people they suddenly understand leverage. 💀

## What I’d actually watch

If you’re considering CoreWeave — or anything in the AI infrastructure theme — keep an eye on:

1. Free cash flow (cash left after all the spending and interest). Positive and growing = the debt story gets easier. Negative and widening = it gets harder.
2. Customer concentration. A handful of giant contracts funding quarter-million-GPU expansion is a fragile setup if one customer rethinks spending.
3. Refinancing. When those bonds mature, what rate does CoreWeave get? That number tells you how the smartest skeptics in the market — bond investors — feel about the company that year.

## The beginner-friendly takeaway

You don’t need to short CoreWeave or buy it to learn from it. The lesson is this: a growing company and a safe company are not the same thing. Growth funded by 15%-yield debt is a bet that demand stays strong long enough for the cash flows to catch up with the obligations. Sometimes that bet pays off spectacularly. Sometimes it doesn’t.

Fun trade and solid business are two different conversations. And “the stock keeps going up” is technically information — it’s just not the information that pays the bondholders. 👀